A Practitioner’s Perspective on Taxonomy, OtlO ntology and ...

A Practitioner’s Perspective on T O t l d Fi d bilitTaxonomy, Ontology and Findability

NASA Johnson Space Center

Sarah BerndtJSC TaxonomistDB Consulting

Information Technology and Multimedia Services ContractMay, 2011

Case Study: Semantic System

This case study follows the NASA “Case of Interest” definition, meaning it illustrates the value of best practices that might otherwise g p gseem insignificant, in order to promote effective implementation.

Product Endorsement DisclaimerThis is not product endorsement but a case study on what we have done and suggestions for the recognition and development of a semantic system.

24/21/2011 JSC Chief Knowledge Officer & Information Resources Directorate Presenter: S. Berndt, ITAMS

Demonstrated Need

Need for broad categories of information that easily map to/ integrate with existing information architectureintegrate with existing information architecture

Need for evolving procedures and processes that are easily maintained and fit users needs

Internal & External analysis Subject Matter Expert (SME) interviews Repository owners, content creators NASA Thesaurus Dow Jones Taxonomy Services Taxonomy consultants


Functional

www.tatapeoplescar.com

JSC needed an

http://www.autoblog.com/media/2009/09/tata-nano-lux-white-630.jpg

uncomplicated, functional system to improve information accessibility and


retrieval.

Succinct Planning = Successful ImplementationSuccessful Implementation

The vision for the JSC Taxonomy is to create a controlled vocabulary to

The envisioned end results are:

The vision for the JSC Taxonomy is to create a controlled vocabulary to connect information stovepipes into an integrated view.

Increased information accessibility, relevancy and currency Improvement of the information consumers’ user

experiencesexperiencesThe scope of the JSC Taxonomy is: To encompass JSC created and/or owned content To include local level vocabularies not to replace existing To include local level vocabularies, not to replace existing

information architecture


Monetized Need

Cost Benefit Analysis

Evidence-based measurement of JSC information workers’Evidence based measurement of JSC information workers search habits, contrasted with industry standards and considering the current search environment/available applications

Information worker = content owners and creators information Information worker = content owners and creators, information consumers

Conservative, salary based value only. Additional costs such as b fit d lti l l ff t t i l d dbenefits and multiple employee efforts not included.

• 8.8 hrs/wk average information worker*

• 10.5 hrs/wk average JSC information worker$21,840 annual cost per employee


* International Data Corporation Hidden Costs of Information Work: A Progress Report, May 2009, Doc # 217936

Investment that Pays

The advantages of planning and technology.

http://www.jaguarxf.info/


Semantic Systems as Tools


System Review

International Standards Organization, ISO 2788 g ,

(establishment of a monolingual thesauri)

to 25964- parts I-IV (thesauri & interoperability with

other vocabularies).

Dublin Core Metadata schemas and vocabulary type

Z39 50 i l f h d i l Z39.50- semantic protocol for search and retrieval

from remote computer databases.


The Advantages of Planning d T h land Technology

http://www.nasa.govSpace Shuttle Discovery, STS-133, on it’s http://www.nasa.gov

final ascent, Febrary 24, 2011

Speeds of over 17,000 mph in ~ 8 minutes = acceleration of 2,000 mph 8 minutes acceleration of 2,000 mph each minute

Two solid rocket boosters

Three Space Shuttle Main Engines (SSMEs)

Two Orbital Maneuvering System (OMS) g y ( )engines to place the vehicle in orbit

38 primary and six vernier Reaction Control System engines for separation


Control System engines for separation and in-space propulsion

Components of the Semantic System

C t ll d Controlled Vocabulary

Hierarchy

Preferred terms

Ontology

Equivalence Relationships

NonPreferred Terms


Components of the Semantic System (2)

Preferred terms to tgenerate

rulebases



Rulebases are i f d b thinformed by the taxonomy and ontology, the proximity andproximity and location of terms, and different weights to enhanceweights to enhance the accuracy of Classification.


Classification Driven Results

Subset of the t tcontent corpus

Fewer, more relevant resultsrelevant results



Term metadata library

Mapping to the interface

Standards for inclusion Standards for inclusion


Term Information Driven Interface

16

Evolution of the Semantic Systemy

Since January 2010, the JSC Taxonomy has more than doubled in term count while the scope of areas covered has grown exponentially! p g p ySeven ‘Top Level’ facets are active in the JSC Google Search.

Production Environment: JSC Taxonomy Term Growth

Within the facets are over 100 ‘Classes’ (major categorizations),

100001200014000160001800020000

PTsRTs

ETs

Total countmany of which are taxonomies in their own right.

Within the classes are over 8 000 f d t ith l

020004000600080008,000 preferred terms with nearly

17,000 relationships betweenthem. These relationships make up the JSC Ontology!gy

Best bet URLs, images and definitions have been added to further enrich results in the Google RTs= Related Terms

PTs= Preferred Terms


gSearch Appliance

PTs Preferred TermsETs= Equivalence Terms

Encourage User Participationg p

Allow for user and Subject Matter Expert participation in the development Allow for user and Subject Matter Expert participation in the development and maintenance of the semantic system.

Identifying the components and functions of a semantic system makes development more palatable Spreadsheets are the most common tools!


development more palatable. Spreadsheets are the most common tools!

Encourage User Participationg p

Consider this a monitoredmonitored folksonomy, meaning the product is centrally b d t tbroadcast, not individually applied.


What We Have Learned

• Read the manual

• Examine the rulebases

• Maintain separate, but integrated domains

• Governance is both overarching and local

D ’ i SME i l h i b ll• Don’t require SMEs to manipulate the semantic system, but allow user participation

• Spread the word• Spread the word

• Test. Test. Test.


What We Suggest gg

• Plan your semantic system based on end user expectations and how the components of your system will meet themhow the components of your system will meet them

• Define the scope for each component and personnel working them

• Consider the content to be represented, are there document management procedures in place to serve as a reference?

Id if h i f b i l d d hi i l i f i i l• Identify the time frame to be included: historical information retrieval, roadmap for the future or both?

• Recognize continuous maintenance and governance needs• Recognize continuous maintenance and governance needs

• Inform the funders, set realistic expectations


Conclusion

At NASA Johnson Space Center (JSC), the Chief Knowledge Officer has b th h i f d l i th JSC T t it li thbeen the champion for developing the JSC Taxonomy to capitalize on the accomplishments of yesterday while maintaining the flexibility needed for the evolving information environment of today.

A clear vision and scope for the semantic system is integral to its success. The vision for the JSC Taxonomy is to connect information stovepipes to present a unified view for information and knowledgestovepipes to present a unified view for information and knowledge across the Center, across organizations, and across decades.

Semantic search at JSC means seamless integration of disparate g pinformation sets into a single interface. Ever increasing use, interest, and organizational participation mark successful integration and provide the framework for future application.


Perspective has its Advantages Too!Tracy Caldwell Dyson

ISS Expedition 24

15.h

tml

gov/

apod

/ap1

011

apod

.nas

a.

A Practitioner’s Perspective on Taxonomy, OtlO ntology and ...

Documents

Transcript of A Practitioner’s Perspective on Taxonomy, OtlO ntology and ...